Posts tagged sound localization
Articles and news from the latest research reports.
Silencers refine sound localization
A new study by LMU researchers shows that sound localization involves a complex interplay between excitatory and inhibitory signals. Pinpointing of sound sources in space would be impossible without the tuning effect of the latter.
Did that lion’s growl come from the left or the right? Or are there two of them out there? In the wild, the ability to perceive sound is of little use unless one can also pinpoint, and discriminate between, different sound sources in space. The capacity for sound localization is equally important for spatial orientation and vocal communication in humans. The underlying mechanism is known to depend on the processing of binaural signals in bilateral nerve-centers in the brainstem, where neural computations extract spatial information is extracted from them. “Each nerve-cell in the processing center receives not only excitatory but also inhibitory signals,” says LMU neurobiologist Professor Benedikt Grothe. “We have now shown how the intrinsic silencing mechanism works at the cellular level, and why it plays such a crucial role in the localization of sounds.”
Sound localization depends on the fact that the “ipsilateral” ear (the one closer to the sound source) perceives the incoming sound slightly earlier than the “contralateral” ear. Since the difference in reception time may be as brief as a fraction of a millisecond, the neural integration process in the time domain must be extremely precise. It was long thought that the direction of the source was determined solely by measuring the difference in the arrival times of excitatory signals from ipsilateral and contralateral ears. But, as Grothe explains: “Comparison of the excitatory signals alone is not sufficient to permit precise discrimination between impulses that arrive only microseconds apart.”
Inhibition reduces background distortion
Using a highly sophisticated experimental design, Grothe and his team were able to demonstrate that spatial information is distilled from four different inputs, namely pairs of inhibitory and excitatory signals arriving from each ear. Moreover, the researchers were able to elucidate the nature of the processing mechanism with the help of a technique known as dynamic patch clamping. With this method, one can measure electrical signals intracellularly, compute their combined effect in real time, and inject the resulting signal back into the cell. “This permits us to measure and manipulate electric currents within cells. By employing this highly complex approach, we were able to characterize the effects of both inhibitory and excitatory signals at the cellular level, and investigate the impact of their integration on the ability to localize sounds,” Grothe explains.
It turns out that neural inhibition controls and dynamically adjusts the time-point at which a given cell becomes maximally active. Thanks to this fine-tuning mechanism, the difference in arrival times between the right and left signals can be determined more precisely than would otherwise be possible. “This is a very dynamic process, which is utilized great precision. Above all, it allows for very rapid resetting of the relationship between the magnitudes of excitatory and inhibitory signals, which would not be feasible on the basis of only two signals,” Grothe adds. How the optimal timing offset is chosen remains unclear, but Grothe hopes that future studies will shed light on this phenomenon.
Listen!
How nerve cells flexibly adapt to acoustic signals: Depending on the input signal, neurons generate action potentials either near or far away from the cell body. This flexibility improves our ability to localize sound sources.
(Image caption: A neuron in the brain stem, that processes acoustic information. Depending on the situation, the cell generates action potentials in the axon (thin process) either close to or far from the body. Photo: Felix Felmy)
In order to process acoustic information with high temporal fidelity, nerve cells may flexibly adapt their mode of operation according to the situation. At low input frequencies, they generate most outgoing action potentials close to the cell body. Following inhibitory or high frequency excitatory signals, the cells produce many action potentials more distantly. This way, they are highly sensitive to the different types of input signals. These findings have been obtained by a research team headed by Professor Christian Leibold, Professor Benedikt Grothe, and Dr. Felix Felmy from the LMU Munich and the Bernstein Center and the Bernstein Focus Neurotechnology in Munich, who used computer models in their study. The researchers report their results in the latest issue of The Journal of Neuroscience.
Did the bang come from ahead or from the right? In order to localize sound sources, nerve cells in the brain stem evaluate the different arrival times of acoustic signals at the two ears. Being able to detect temporal discrepancies of up to 10 millionths of a second, the neurons have to become excited very quickly. In this process, they change the electrical voltage that prevails on their cell membrane. If a certain threshold is exceeded, the neurons generate a strong electrical signal — a so-called action potential — which can be transmitted efficiently over long axon distances without weakening. In order to reach the threshold, the input signals are summed up. This is achieved easier, the slower the nerve cells alter their electrical membrane potential.
Input signals are optimally processed
These requirements — rapid voltage changes for a high temporal resolution of the input signals, and slow voltage changes for an optimal signal integration that is necessary for the generation of an action potential — represent a paradoxical challenge for the nerve cell. “This problem is solved by nature by spatially separating the two processes. While input signals are processed in the cell body and the dendrites, action potentials are generated in the axon, a cell process,” says Leibold, leader of the study. But how sustainable is the spatial separation?
In their study, the researchers measured the axons’ geometry and the threshold of the corresponding cells and then constructed a computer model that allowed them to investigate the effectiveness of this spatial separation. The researchers’ model predicts that depending on the situation, neurons produce action potentials with more or less proximity to the cell body. For high frequency or inhibitory input signals, the cells will shift the location from the axon’s starting point to more distant regions. In this way, the nerve cells ensure that the various kinds of input signals are optimally processed — and thus allow us to perceive both small and large acoustic arrival time differences well, and thereby localize sounds in space.
(Source: en.uni-muenchen.de)
Researchers uncover why there is a mapping between pitch and elevation
Have you ever wondered why most natural languages invariably use the same spatial attributes – high versus low – to describe auditory pitch? Or why, throughout the history of musical notation, high notes have been represented high on the staff? According to a team of neuroscientists from Bielefeld University, the Max Planck Institute for Biological Cybernetics in Tübingen and the Bernstein Center Tübingen, high pitched sounds feel ‘high’ because, in our daily lives, sounds coming from high elevations are indeed more likely to be higher in pitch. This study has just appeared in the science journal PNAS.
Dr. Cesare Parise and colleagues set out to investigate the origins of the mapping between sound frequency and spatial elevation by combining three separate lines of evidence. First of all, they recorded and analyzed a large sample of sounds from the natural environment and found that high frequency sounds are more likely to originate from high positions in space. Next, they analyzed the filtering of the human outer ear and found that, due to the convoluted shape of the outer ear – the pinna – sounds coming from high positions in space are filtered in such a way that more energy remains for higher pitched sounds. Finally, they asked humans in a behavioural experiment to localize sounds with different frequency and found that high frequency sounds were systematically perceived as coming from higher positions in space.
The results from these three lines of evidence were highly convergent, suggesting that all such diverse phenomena as the acoustics of the human ear, the universal use of spatial terms for describing pitch, or the reason why high notes are represented higher in musical notation ultimately reflect the adaptation of human hearing to the statistics of natural auditory scenes. ‘These results are especially fascinating, because they do not just explain the origin of the mapping between frequency and elevation,’ says Parise, ‘they also suggest that the very shape of the human ear might have evolved to mirror the acoustic properties of the natural environment. What is more, these findings are highly applicable and provide valuable guidelines for using pitch to develop more effective 3D audio technologies, such as sonification-based sensory substitution devices, sensory prostheses, and more immersive virtual auditory environments.’
The mapping between pitch and elevation has often been considered to be metaphorical, and cross-sensory correspondences have been theorized to be the basis for language development. The present findings demonstrate that, at least in the case of the mapping between pitch and elevation, such a metaphorical mapping is indeed embodied and based on the statistics of the environment, hence raising the intriguing hypothesis that language itself might have been influenced by a set of statistical mappings between naturally occurring sensory signals.
Besides the mapping between pitch and elevation, human perception, cognition, and action are laced with seemingly arbitrary correspondences, such as that yellow–reddish colors are associated with a warm temperature or that sour foods taste sharp. This study suggests that many of these seemingly arbitrary mappings might in fact reflect statistical regularities to be found in the natural environment.
Researchers Turn Current Sound-localization Theories ‘On their Ear’
The ability to localize the source of sound is important for navigating the world and for listening in noisy environments like restaurants, an action that is particularly difficult for elderly or hearing impaired people. Having two ears allows animals to localize the source of a sound. For example, barn owls can snatch their prey in complete darkness by relying on sound alone. It has been known for a long time that this ability depends on tiny differences in the sounds that arrive at each ear, including differences in the time of arrival: in humans, for example, sound will arrive at the ear closer to the source up to half a millisecond earlier than it arrives at the other ear. These differences are called interaural time differences. However, the way that the brain processes this information to figure out where the sound came from has been the source of much debate.
A recent paper by Mass. Eye and Ear/Harvard Medical School researchers in collaboration with researchers at the Ecole Normale Superieure, France, challenge the two dominant theories of how people localize sounds, explain why neuronal responses to sounds are so diverse and show how sound can be localized, even with the absence of one half of the brain. Their research is described on line in the journal eLife.
“Progress has been made in laboratory settings to understand how sound localization works, but in the real world people hear a wide range of sounds with background noise and reflections,” said Dan F. M. Goodman, lead author and post-doctoral fellow in the Eaton-Peabody Laboratories at Mass. Eye and Ear, Harvard Medical School. “Theories based on more realistic environments are important. The theme of the paper is that previous theories about this have been too idealized, and if you use more realistic data, you come to an entirely different conclusion.”
“Two theories have come to dominate our understanding of how the brain localizes sounds: the peak coding theory (which says that only the most strongly responding brain cells are needed), and the hemispheric coding theory (which says that only the average response of the cells in the two hemispheres of the brain are needed),” Goodman said. “What we’ve shown in this study is that neither of these theories can be right, and that the evidence they presented only works because their experiments used unnatural/idealized sounds. If you use more realistic, natural sounds, then they both do very badly at explaining the data.”
Researchers showed that to do well with realistic sounds, one needs to use the whole pattern of neural responses, not just the most strongly responding or average response. They showed two other key things: first, it has long been known that the responses of different auditory neurons are very diverse, but this diversity was not used in the hemispheric coding theory.
“We showed that the diversity is essential to the brain’s ability to localize sounds; if you make all the responses similar then there isn’t enough information, something that was not appreciated before because if one has unnatural/idealized sounds you don’t see the difference” Goodman said.
Second, previous theories are inconsistent with the well-known fact that people are still able to localize sounds if they lose one half of our brain, but only sounds on the other side (i.e. if one loses the left half of the brain, he or she can still localize sounds coming from the right), he added.
“We can explain why this is the case with our new theory,” Goodman said.
(Source: masseyeandear.org)